Temperature and precipitation change over South China in CMIP5 and CMIP6 models: historical simulation and future projection

Revealing regional climate changes is vital for policymaking activities related to climate change adaptation and mitigation. South China is a well-developed region with dense population, but the climate projection uncertainty remains unevaluated in detail. Here, we comprehensively assess the historical simulations and future projection of climate change in South China based on the CMIP5/CMIP6 models. We show evidence that CMIP5/CMIP6 models can well reproduce the observed distributions of annual/seasonal mean temperature but show much lower skills for the precipitation. CMIP6 outperform CMIP5 in the historical simulations, as evidenced by more models with lower bias magnitude and higher skill scores. During 2021-2100, the annual mean temperature over South China would increase significantly at a rate (℃ decade-1) of 0.53 (0.42~0.63) and 0.59 (0.52~0.66), while the precipitation would increase slightly at a rate (% decade-1) of 0.78 (0.15~1.56) and 1.52 (0.91~2.30), under RCP8.5 and SSP5-8.5 scenarios, respectively. CMIP6 models project larger annual/seasonal mean temperature and precipitation trends than CMIP5 models under the equivalent scenarios. South China would robustly increase by more than 1.5 ℃ during 2041-2060 under RCP4.5 and SSP2-4.5 while 4.5 ℃ during 2081-2100 under RCP8.5 and SSP5-8.5 with respect to 1850-1900. The temperature projection uncertainty is mainly dominated by model uncertainty and scenario uncertainty, while internal uncertainty contributes some during the near-term. The precipitation projection uncertainty is mainly from internal uncertainty and model uncertainty. For both temperature and precipitation projection uncertainty, the relative sizes of contributions from the main contributors vary in time and show obvious seasonal differences.